Topping PA5 Review (Amplifier)

[Klint]

The PA5 has only a single shared 2200uF of bulk rail capacitance between the PSU and the TPA3255 chip. The value recommended by the TI reference design is two 4700uF (one on each channel). This would dramatically increase the charge reserves for deep bass transients and also improve stereo channel separation. This is actually one area that was "skimped" probably due to the physical package limitations of making it fit into the standard box size shared with the L50 that it was meant to be used with. On my amp, I have dual 63v rated Nichicon 4700uF rail caps. I also have the +ve pin of the cap mounted only 15mm away from the Vdd power input pins of the TPA3255 using 2oz copper. Little details like this can matter, as well as using high strand count low impedance silicone "RC drone motor/battery" wire between the PSU and the amp and the amp and the binding posts.
View attachment 171621

Now im having second thoughts, i was happy and unknowing. I do know i have 60 000 in my current amp, A/B. How much is good? Thanks for bringing it up.

 

[restorer-john]

Now im having second thoughts, i was happy and unknowing. I do know i have 60 000 in my current amp, A/B. How much is good? Thanks for bringing it up.

You can't compare the PSU filter capacitances in Class A/AB amplifier with conventional transformer based supplies to an external high frequency SMPS feeding a Class D with some localised (to the amp board) reservior capacitance.

That 2200uF capacitance is not the main supply capacitance- it is there to keep the supply impedance low locally. The main SMPS will have more than enough capacitance to give a super low amount of ripple and low source impedance. Consider it is being charged at many kilohertz, perhaps 50-100kHz or greater, not only 50/60Hz like a typical supply.

The choice of that capacitor no doubt also takes into consideration the inductance of the SMPS supply cable to the amplifier. Make it too big and the overall supply performance would suffer.

 

[JohnYang1997]

My PA5 just arrived in the mail.

Sadly the speaker output jacks are loose and flopping around as others have mentioned - and the manual makes it very clear that the warranty is "instantly" voided if you open it.

@JohnYang1997 Is it OK to open it up ourselves to repair this issue without voiding the warranty?

It's ok to do so.

 

[JohnYang1997]

The PA5 has only a single shared 2200uF of bulk rail capacitance between the PSU and the TPA3255 chip. The value recommended by the TI reference design is two 4700uF (one on each channel). This would dramatically increase the charge reserves for deep bass transients and also improve stereo channel separation. This is actually one area that was "skimped" probably due to the physical package limitations of making it fit into the standard box size shared with the L50 that it was meant to be used with. On my amp, I have dual 63v rated Nichicon 4700uF rail caps. I also have the +ve pin of the cap mounted only 15mm away from the Vdd power input pins of the TPA3255 using 2oz copper. Little details like this can matter, as well as using high strand count low impedance silicone "RC drone motor/battery" wire between the PSU and the amp and the amp and the binding posts.
View attachment 171621

20hz power graph and crosstalk measurements showed otherwise. Adding more doesn't do any better especially at 38V.

 

[JohnYang1997]

If you do drive a Class D amp at higher power and leave the output unconnected, it will destroy the amp chip (blow up in spectacular flames). This is because the back EMF generated by the inductor during the switching has to go somewhere, and the voltage can shoot well beyond what the internal MOSFETs can handle. I have had this happen under max power stress testing when I forgot to disconnect the input to one of the channels during the test as I only had a fan cooled 300W load resistor on the output of the other channel.

Some SMT rework to remove the dead TPA3255 and replace it worked well but it is not a job for every DIY’er.

This is not exactly true. There should be no issue testing unloaded to from low voltage to clipping. If your amp has that behavior then something needs to be altered in the design, eg snubber and/or EMI filter values.
Disconnecting the load while testing at high power into load may cause issue which I have not tried.

 

[JohnYang1997]

Also to those of you who are convinced PA5 used certain chip, you can hot air replace one onto it and see if it functions normally. It can not would not.
And multiple cares are taken into making the clipping behavior good, frequency response +-0.3dB from 10-40khz (you won't find this elsewhere) and already known low noise and distortion. You can't just increase feedback and call it the day, it's only the start.
Also like mentioned, it took some serious work to make things work. I have developed accurate simulation models and novel optimization methods. You are not finding these anywhere else.
After my NFCA circuit got copied and supposedly patented(in China) by a different company and even there are clear errors in the patent filing, it drove me mad like no tomorrow. I of course would go extra mile protecting all the hours that went in.
And to people who think it's still overpriced. This is sold globally with many authorized resellers with hierarchy. The provided power supply is actually 300W supply. All of this I don't think you'll get anything better for cheaper.

 

[xrk971]

Hi John Yang,
Your work is to be commended - and I for one, am
very impressed by the performance and price point. I do not think it’s overpriced at all. My hat’s off to you.

That’s interesting that you say it’s not a normal run of the mill chip we can swap to do this. Well lots more head scratching on this then, but armchair reverse engineering costs nothing, so nothing lost.

Thank you for your development of this product. Very nice.

 

[Kevinfc]

Also to those of you who are convinced PA5 used certain chip, you can hot air replace one onto it and see if it functions normally. It can not would not.
And multiple cares are taken into making the clipping behavior good, frequency response +-0.3dB from 10-40khz (you won't find this elsewhere) and already known low noise and distortion. You can't just increase feedback and call it the day, it's only the start.
Also like mentioned, it took some serious work to make things work. I have developed accurate simulation models and novel optimization methods. You are not finding these anywhere else.
After my NFCA circuit got copied and supposedly patented(in China) by a different company and even there are clear errors in the patent filing, it drove me mad like no tomorrow. I of course would go extra mile protecting all the hours that went in.
And to people who think it's still overpriced. This is sold globally with many authorized resellers with hierarchy. The provided power supply is actually 300W supply. All of this I don't think you'll get anything better for cheaper.

Thanks for coming around and putting your thoughts out. Sometimes it can be a little hostile around here, but I think I speak for most that it is appreciate.

 

[xrk971]

This is not exactly true. There should be no issue testing unloaded to from low voltage to clipping. If your amp has that behavior then something needs to be altered in the design, eg snubber and/or EMI filter values.
Disconnecting the load while testing at high power into load may cause issue which I have not tried.

This happened to a TPA3255 implemented following the TI Reference design as to LC filter and snubber. It only blew up when I ran at about 125w for over 30 seconds without a load. This was a stress test on a 300w load resistor. The channel without the resistor blew up. I don’t think it would have blown up at 75w. The LC filter is designed assuming a particular load to damp the oscillations. I suppose catch diodes had they been implemented would have e prevented this.

 

[AudioArchitech]

Very tempted to sell my Audiolab 6000A for this Topping PA5 and put the difference in cash towards a MiniDSP 2x4HD. The crossover will reduce the load to PA5 to my mains, sub for rest.

 

[pma]

It only blew up when I ran at about 125w for over 30 seconds without a load.

Maybe better to tell the output voltage before load was disconnected? “125W” ”without a load” does not make much sense.

 

[xrk971]

I should clarify. One side had a load and other side empty no load. But both inputs were driven by same signal. I forgot that both channel inputs were connected but a load was only connected on one side. The side without the load blew up.

 

[envydd]

Also to those of you who are convinced PA5 used certain chip, you can hot air replace one onto it and see if it functions normally. It can not would not.
And multiple cares are taken into making the clipping behavior good, frequency response +-0.3dB from 10-40khz (you won't find this elsewhere) and already known low noise and distortion. You can't just increase feedback and call it the day, it's only the start.
Also like mentioned, it took some serious work to make things work. I have developed accurate simulation models and novel optimization methods. You are not finding these anywhere else.
After my NFCA circuit got copied and supposedly patented(in China) by a different company and even there are clear errors in the patent filing, it drove me mad like no tomorrow. I of course would go extra mile protecting all the hours that went in.
And to people who think it's still overpriced. This is sold globally with many authorized resellers with hierarchy. The provided power supply is actually 300W supply. All of this I don't think you'll get anything better for cheaper.

congrats. I ordered one to support the effort. Hope we see even better designs in the future.

 

[envydd]

@JohnYang1997 whats a good dac to go with the pa5? I currently have a smsl m200 dac. I prefer the volume control knob on the dac else would have ordered the e50.

 

[DanielT]

Very tempted to sell my Audiolab 6000A for this Topping PA5 and put the difference in cash towards a MiniDSP 2x4HD. The crossover will reduce the load to PA5 to my mains, sub for rest.

I think that sounds interesting. PA5, no matter how good it is, has too little effect for me to have in the main system in the living room. But with HP-LP filter, miniDSP, sub. Hm... hmmmm.

In which case, well done Topping and @JohnYang1997 with PA5!

 

[AudioArchitech]

Curious to hear this reply as well. My intuition tells me that the D30 Pro could have great synergy

 

[mdsimon2]

Very tempted to sell my Audiolab 6000A for this Topping PA5 and put the difference in cash towards a MiniDSP 2x4HD. The crossover will reduce the load to PA5 to my mains, sub for rest.

Only issue I see with the 2X4HD is that with 2 V output you will not be able drive the PA5 to full power and it is not exactly super high powered to begin with. Although the relatively low gain of 19 dB on the PA5 would definitely help with hiss as the 2X4HD is rather noisy.

Michael

 

[F1308]

After my NFCA circuit got copied and supposedly patented(in China) by a different company and even there are clear errors in the patent filing, it drove me mad like no tomorrow.

I am very sorrow and feel pain hearing this.
My deepest condolences and best of understandings.
Also, my 120 minutes of applause for your achievements.
Well done.
Great work.
Lovely.
Thank you very much, indeed.

I have been on standby waiting to buy as I went following your comments on the works you were developing and announced hereabouts with a whisper, and finally did buy.

And now, tell us, please, the name of the villain so I, we, can possibly avoid becoming his customers.

 

[AudioArchitech]

I'll share my workaround on that as soon as I can. Want to get my hands on it first before sharing : )

 

[voodooless]

Also to those of you who are convinced PA5 used certain chip, you can hot air replace one onto it and see if it functions normally. It can not would not.

If I were you (and Topping), I would have kept us in that delusion. It would have brought multiple additional days of sniggering and laughter to you guys.

On the other hand, now you can see us endulge in another round of conspiracies . I guess either way, you win